Artificial intelligence could become a threat not only to jobs and businesses but to the stability of the global financial system. Bank of England governor Andrew Bailey has warned G20 policymakers that increasingly capable frontier AI models could intensify cyberattacks and allow disruption to spread rapidly between countries. His concerns come as financial authorities also confront stretched market valuations and growing investor enthusiasm around AI.
Cyber risk grows
Writing to G20 finance ministers and central bank governors ahead of their meeting in North Carolina, Bailey said frontier models are developing more sophisticated autonomy, problem-solving skills and potentially dangerous capabilities. Many countries, he warned, still lack adequate protocols governing how such systems are developed, released and deployed.
Financial markets could be particularly vulnerable because banks, exchanges and other institutions operate through highly interconnected systems. Advanced AI may change the speed, scale and cost of cyberattacks, allowing a serious incident to move across borders and undermine confidence throughout the system.
Concentrated reliance on third-party technology providers adds another weakness. A successful attack affecting one widely used provider could consequently cause disruption across numerous financial institutions at once. Bailey, who also chairs the Financial Stability Board, called for coordinated international action supporting the safe and responsible release of advanced models.
Market correction fears grow
Cybersecurity is not his only concern. High valuations, growing leverage in bond and equity markets and concentrated investor bets could combine to make the financial system more vulnerable to a sharp correction. Much of that optimism has been fuelled by expectations surrounding AI. Bailey warned that a major shock, or several shocks arriving together, could trigger multiple financial vulnerabilities simultaneously.
I remain concerned therefore that a large shock or combination of shocks could concurrently trigger multiple vulnerabilities,
Andrew Bailey
His comments follow broader warnings from people developing advanced AI themselves. Last month, 1,367 researchers and engineers from frontier laboratories including OpenAI, Anthropic and Google DeepMind signed a letter warning that AI capabilities could accelerate faster than humanity’s ability to understand or control them.
With increasingly powerful models entering an already interconnected financial system, Bailey wants regulators to treat AI risk as an international problem rather than one individual countries can contain alone.
